match-standards-water-mix04.Rmd

---
title: "Identifying mix04 standards in water"
author: "Andrea Vicini, Vinicius Verri Hernandes, Johannes Rainer"
output:
  rmarkdown::html_document:
    highlight: pygments
    toc: true
    toc_float: true
    toc_depth: 3
    fig_width: 5
---

```{r, include = FALSE, cached = FALSE}
knitr::read_chunk("R/match-standards-chunks.R")
MIX <- 4
MATRIX <- "Water"
POLARITY <- "POS"
## settings to find MS2 spectra for features. Be rather inclusive here.
FEATURE_MS2_PPM <- 20
FEATURE_MS2_TOLERANCE <- 0.05
```

```{r, libraries, echo = FALSE, message = FALSE}
```
```{r, general-settings, echo = FALSE, message = FALSE}
```

**Current version: 0.10.0, 2022-05-30.**

# Introduction

Mixes of standards have been solved in water or added to human serum sample
pools in two different concentration and these samples were measured with the
LC-MS setup from Eurac (used also to generate the CHRIS untargeted metabolomics
data). Assignment of retention time and observed ion can be performed for each
standard based on MS1 information such as expected m/z but also based on the
expected difference in signal between samples with low and high concentration of
the standards. Finally, experimental fragment spectra provide the third level of
evidence. Thus, the present data set allows to annotate features to standards
based on 3 levels of evidence: expected m/z, difference in measured intensity
and MS2-based annotation.

A detailed description of the approach is given in
[match-standards-introduction.Rmd](match-standards-introduction.Rmd).


The list of standards that constitute the present sample mix are listed below
along with the expected retention time and the most abundant adduct for positive
and negative polarity as defined manually in a previous analysis by Mar
Garcia-Aloy.

```{r, standards-table, echo = FALSE, results = "asis"}
```

# Data import

We next load the data and split it according to matrix and polarity.

```{r, data-import}
```

We next also load the reference databases we will use to compare the
experimental MS2 spectra against. Below we thus load HMDB and MassBank data. We
create in addition *neutral loss* versions of the databases. Since HMDB does not
provide precursor m/z values we assume the fragment spectra to represent `[M+H]+`
ions (respectively `[M-H]-` ions for negative polarity) and add the m/z for
these adducts as *precursor m/z*.

```{r, load-reference-databases, message = FALSE}
```

```{r, setup-ion-db, message = FALSE, echo = FALSE}
```

# Initial evaluation of available MS2 data

Before performing the actual analysis that involves chromatographic peak
detection and evaluation of feature abundances, we compare all experimental MS2
spectra against the reference libraries. This provides already a first hint for
which standards a valid MS2 spectrum was recorded (and hence would allow
annotation based on evidence 3) and what related retention time might be.

We first extract all MS2 spectra from the respective data files and process them
by first removing all peaks with an intensity below 5% of the highest peak
signal, then scaling all intensities to a value range between 0 and 100 and
finally remove all MS2 spectra with less than 2 peaks.

```{r prepare-all-ms2}
fls <- fileNames(data_all)[data_all$polarity == "POS"]
```
```{r, all-ms2}
```

We next match these spectra against all reference spectra from HMDB for the
standards in `r MIX_NAME`. 

The settings for the matching are defined below. These will be used for all MS2
spectra similarity calculations. All matching spectra with a similarity larger
0.7 are identified.

```{r, compare-spectra-param}
```
 
Standards for which no MS2 spectrum in HMDB is present are listed in
the table below.

```{r, echo = FALSE, results = "asis"}
tmp <- std_dilution[!(std_dilution$HMDB %in% hmdb_std$compound_id), ]
pandoc.table(tmp[, c("name", "HMDB", "formula")], style = "rmarkdown",
             caption = "Standards for which no reference spectrum is available")
```

The table below lists all standards and the number of matching reference spectra
(along with their retention times).

```{r, table-all-ms2-hmdb, echo = FALSE, results = "asis"}
```

We repeat the same analysis for the negative polarity data (code not shown
again).

```{r, echo = FALSE}
fls <- fileNames(data_all)[data_all$polarity == "NEG"]
```
```{r, all-ms2, echo = FALSE}
```
```{r, table-all-ms2-hmdb, echo = FALSE, results = "asis"}
```

We next perform the full analysis of the data set involving MS1 and MS2 data.


# Water, positive polarity

We first perform the analysis on the samples with the standards solved in pure
water acquired in positive polarity mode.

```{r}
POLARITY <- "POS"
data <- filterFile(data_all, which(data_all$polarity == POLARITY))
hmdb <- hmdb_pos
hmdb_nl <- hmdb_pos_nl
```

## Data pre-processing

We perform the pre-processing of our data set which consists of the
chromatographic peak detection followed by a peak refinement step to reduce peak
detection artifacts, the correspondence analysis to group peaks across samples
and finally the gap-filling to fill in missing peak data from samples in which
no chromatographic peak was detected. For details on settings please refer to
the analysis of *mix 01* samples.

```{r, preprocessing, eval = !file.exists(paste0(RDATA_PATH, "processed_data_", POLARITY, ".RData")), message = FALSE}
```

```{r, echo = FALSE}
load(paste0(RDATA_PATH, paste0("processed_data_", POLARITY, ".RData")))
data_FS <- filterFile(data, which(data$mode == "FS"),
                      keepFeatures = TRUE)
```

## Signal intensity difference

We compute the difference in (log2) signals between samples with high and low
concentration of the standards and calculate the p-value for this difference
using the Student's t-test.

```{r, abundance-difference}
```

## Identification of features matching standards

We now identify all features matching the any of the pre-defined set of adducts
for the standards of mix `r MIX`. Matching features are further subsetted to
those that show an at least twice as high signal in samples with high
concentration compared to those with low concentrations. Also, features with a
signal in low concentration but no detectable signal in high concentration are
removed.

```{r, define-adducts-pos}
```
```{r, match-features}
```

For most of the standards (`r length(unique(mD$target_name))` out of 
`r nrow(std_dilution)`) in mix `r MIX` at least one feature was found
matching the standards adducts' m/z. 

```{r, table-standard-no-feature, results = "asis", echo = FALSE}
```

In the next sections we investigate for each standard which assignment would be
the correct one or, for those for which no signal was detected, why that was the
case.

## Standards with matching features

While for some standards a matching feature was found we still need to evaluate
whether this matching is correct. For each standard we thus first evaluate the
EICs for all matching features, then we match their MS2 spectra (if available)
against reference libraries. To ensure correct assignment of a feature, its
retention time and eventually related MS2 spectra, we consider the following
criteria to determine the annotation confidence:

- feature(s) was/were found with m/z matching those of adduct(s) of the 
  standard.
- signal is higher for samples with higher concentration.
- MS2 spectra matches reference spectra for the standard (if available).
- MS2 spectra don't match MS2 spectra of other compounds.

### Acetylalanine

```{r, echo = FALSE}
std <- "Acetylalanine"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

Many features have been assigned to this standard based on
m/z matching. Based on their retention times, these seem however to represent
ions from different compounds.

We next plot the EICs for the assigned features and visually inspect these.

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WP", std_ms2)
```

The cleaned MS2 spectra for the features matched to `r std` are shown below
(peaks with an intensity below 5% of the maximum peak and spectra with less
than 2 peaks were removed).

```{r mix04-water-pos-acetylalanine-ms2, echo = FALSE}
plot_spectra(std_ms2)
```

We next match the extracted MS2 spectra against the reference spectra for 
`r std` from HMDB and MassBank.

```{r mix04-water-pos-acetylalanine-ms2-hmdb, echo = FALSE, fig.width = 5, fig.height = 5}
hmdb_id <- std_dilution$HMDB[std_dilution$name == std]
std_hmdb <- Spectra(cdb, filter = ~ compound_id == hmdb_id)
sim <- plot_ms2_similarity_heatmap(std_ms2, std_hmdb, csp)
```

```{r mix04-water-pos-acetylalanine-acid-ms2-mbank, echo = FALSE, fig.width = 5, fig.height = 5}
std_mbank <- get_mbank(mbank, inchikey = unique(std_hmdb$inchikey))
sim <- plot_ms2_similarity_heatmap(std_ms2, std_mbank, csp)
```

In addition we perform a neutral loss spectra comparison.

```{r mix04-water-pos-acetylalanine-ms2-hmdb_nl, echo = FALSE, fig.width = 5, fig.height = 5}
std_hmdb_nl <- hmdb_nl[hmdb_nl$compound_id == hmdb_id]
std_ms2_nl <- neutralLoss(std_ms2, nl_param)
sim <- plot_ms2_similarity_heatmap(std_ms2_nl, std_hmdb_nl, csp_nl)
```

```{r mix04-water-pos-acetylalanine-hmdb-nl, echo = TRUE, fig.cap = "Mirror plots"}
tmp_sel <- plot_select_ms2(std_ms2_nl, std_hmdb_nl, 0.2, ppm = csp_nl@ppm,
                           tolerance = csp_nl@tolerance)
```

Neutral loss spectra seem to match to some degree.

```{r mix04-water-pos-acetylalanine-ms2-mbank_nl, echo = FALSE, fig.width = 5, fig.height = 5}
std_mbank_nl <- get_mbank(mbank, inchikey = unique(std_hmdb$inchikey), nl = TRUE)
std_ms2_nl <- neutralLoss(std_ms2, nl_param)
sim <- plot_ms2_similarity_heatmap(std_ms2_nl, std_mbank_nl, csp_nl)
```

In addition we match (**all**) the MS2 spectra for the matched features against
all spectra from HMDB or MassBank identifying reference spectra with a
similarity larger than 0.7. The results (if any spectra matched) are shown in
the two following tables.

```{r, echo = FALSE, message = FALSE, results = "asis"}
std_ms2_sel <- Spectra()
perform_match(std_ms2, hmdb, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, mbank, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```


#### Summary

- Inconclusive. `[M+Na]+` seems to be the *closest* real signal, also based on
  the EIC. Needs to be confirmed for negative polarity and serum.
- Annotation of FT01376 (RT=65, `[M+Na]+` ion) to `r std` with confidence level
  **C**? (because of neutral loss spectra comparison)
- Reference MS2: FT01376 `[M+Na]+`: FT01376_F07.S0310, FT01376_F08.S0292
  (low confidence).


### Allantoin

```{r, echo = FALSE}
std <- "Allantoin"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

Many features have been assigned to this standard based on
m/z matching. Based on their retention times, these seem however to represent
ions from different compounds.

We next plot the EIC for the assigned feature and visually inspect these.

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WP", std_ms2)
```

The cleaned MS2 spectra for the features matched to `r std` are shown below
(peaks with an intensity below 5% of the maximum peak and spectra with less
than 2 peaks were removed).

```{r mix04-water-pos-allantoin-ms2, echo = FALSE}
plot_spectra(std_ms2)
```

We next match the extracted MS2 spectra against the reference spectra for 
`r std` from HMDB and MassBank.

```{r mix04-water-pos-allantoin-ms2-hmdb, echo = FALSE, fig.width = 5, fig.height = 5}
hmdb_id <- std_dilution$HMDB[std_dilution$name == std]
std_hmdb <- Spectra(cdb, filter = ~ compound_id == hmdb_id)
sim <- plot_ms2_similarity_heatmap(std_ms2, std_hmdb, csp)
```

```{r mix04-water-pos-allantoin-acid-ms2-mbank, echo = FALSE, fig.width = 5, fig.height = 5}
std_mbank <- get_mbank(mbank, inchikey = unique(std_hmdb$inchikey))
sim <- plot_ms2_similarity_heatmap(std_ms2, std_mbank, csp)
```

In both cases we don't find any match. In addition we perform a neutral loss
spectra comparison.

```{r mix04-water-pos-allantoin-ms2-hmdb_nl, echo = FALSE, fig.width = 5, fig.height = 5}
std_hmdb_nl <- hmdb_nl[hmdb_nl$compound_id == hmdb_id]
std_ms2_nl <- neutralLoss(std_ms2, nl_param)
sim <- plot_ms2_similarity_heatmap(std_ms2_nl, std_hmdb_nl, csp_nl)
```

```{r mix04-water-pos-allantoin-ms2-mbank_nl, echo = FALSE, fig.width = 5, fig.height = 5}
std_mbank_nl <- get_mbank(mbank, inchikey = unique(std_hmdb$inchikey), nl = TRUE)
std_ms2_nl <- neutralLoss(std_ms2, nl_param)
sim <- plot_ms2_similarity_heatmap(std_ms2_nl, std_mbank_nl, csp_nl)
```

For the HMDB comparison we don't find matches while for the MassBank one
the neutral loss spectra cannot be computed. 

In addition we match (**all**) the MS2 spectra for the matched features against
all spectra from HMDB or MassBank identifying reference spectra with a
similarity larger than 0.7. The results (if any spectra matched) are shown in
the two following tables.

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, hmdb, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

There are high similarity matches involving compounds different from `r std`.

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, mbank, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```


#### Summary

- Inconclusive. MS2 spectra don't match references.
- EICs also don't look that good.


### Gamma-Aminobutyric acid

```{r, echo = FALSE}
std <- "Gamma-Aminobutyric acid"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

Many features have been assigned to this standard based on
m/z matching. Based on their retention times, these seem however to represent
ions from different compounds.

We next plot the EIC for the assigned feature and visually inspect these.

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WP", std_ms2)
```

The cleaned MS2 spectra for the features matched to `r std` are shown below
(peaks with an intensity below 5% of the maximum peak and spectra with less
than 2 peaks were removed).

```{r mix04-water-pos-gamma-aminobutyric-acid-ms2, echo = FALSE}
plot_spectra(std_ms2)
```

We next match the extracted MS2 spectra against the reference spectra for 
`r std` from HMDB and MassBank.

```{r mix04-water-pos-gamma-aminobutyric-acid-ms2-hmdb, echo = FALSE, fig.width = 5, fig.height = 5}
hmdb_id <- std_dilution$HMDB[std_dilution$name == std]
std_hmdb <- Spectra(cdb, filter = ~ compound_id == hmdb_id)
sim <- plot_ms2_similarity_heatmap(std_ms2, std_hmdb, csp)
```

```{r mix04-water-pos-gamma-aminobutyric-acid-mirror-hmdb, echo = TRUE, fig.cap = "Mirror plots"}
std_ms2_sel <- plot_select_ms2(std_ms2, std_hmdb, 0.7,
                               ppm = csp@ppm, tolerance = csp@tolerance)
```

```{r mix04-water-pos-gamma-aminobutyric-acid-acid-ms2-mbank, echo = FALSE, fig.width = 5, fig.height = 5}
std_mbank <- get_mbank(mbank, inchikey = unique(std_hmdb$inchikey))
sim <- plot_ms2_similarity_heatmap(std_ms2, std_mbank, csp)
```

```{r mix04-water-pos-gamma-aminobutyric-acid-mirror-mbank, echo = TRUE, fig.cap = "Mirror plots"}
tmp_sel <- plot_select_ms2(std_ms2, std_mbank, 0.7,
                           ppm = csp@ppm, tolerance = csp@tolerance)
```

In addition we match (**all**) the MS2 spectra for the matched features against
all spectra from HMDB or MassBank identifying reference spectra with a
similarity larger than 0.7. The results (if any spectra matched) are shown in
the two following tables.

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, hmdb, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, mbank, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

#### Summary

- FT00438 (RT = 134.2, `[M+H]+` ion) with confidence **B**.
- FT00214 (RT = 134.3, `[M+H-NH3]+` ion) with confidence **B-**.
- FT00203 (RT = 133.9, `[M+H-H2O]+` ion) with confidence **B-**.
- Reference MS2: `[M+H]+`: FT00438_F07.S0662, FT00438_F08.S0654; high
  confidence. `[M+H-NH3]+`: FT00214_F07.S0641, FT00214_F07.S0717,
  FT00214_F08.S0565, FT00214_F08.S0633; low confidence. `[M+H-H2O]+`:
  FT00203_F07.S0584, FT00203_F07.S0647, FT00203_F08.S0582, FT00203_F08.S0676;
  low confidence.

```{r, echo = FALSE}
## Get MS2 spectra:
ms2 <- std_ms2[c("FT00438_F07.S0662", "FT00438_F08.S0654", "FT00214_F07.S0641",
                 "FT00214_F07.S0717", "FT00214_F08.S0565", "FT00214_F08.S0633",
                 "FT00203_F07.S0584", "FT00203_F07.S0647", "FT00203_F08.S0582",
                 "FT00203_F08.S0676")]
ms2$confidence <- c(rep("high", 2), rep("low", 8))
```

```{r, add-ms2-spectra, echo = FALSE}
```


### AMP

```{r, echo = FALSE}
std <- "AMP"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

Many features have been assigned to this standard based on m/z matching.
We next plot their EIC and visually inspect them.

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WP", std_ms2)
```

The extracted MS2 spectra for the features matched to `r std` is shown
below (peaks with an intensity below 5% of the maximum peak where removed).

```{r mix04-water-pos-amp-ms2, echo = FALSE}
plot_spectra(std_ms2)
```

We next match the extracted MS2 spectrum against the reference spectra for 
`r std` from HMDB and MassBank.

```{r mix04-water-pos-amp-ms2-hmdb, echo = FALSE, fig.width = 5, fig.height = 5}
hmdb_id <- std_dilution$HMDB[std_dilution$name == std]
std_hmdb <- Spectra(cdb, filter = ~ compound_id == hmdb_id)
sim <- plot_ms2_similarity_heatmap(std_ms2, std_hmdb, csp)
```

```{r mix04-water-pos-amp-mirror-hmdb, echo = TRUE, fig.cap = "Mirror plots"}
std_ms2_sel <- plot_select_ms2(std_ms2, std_hmdb, 0.7,
                               ppm = csp@ppm, tolerance = csp@tolerance)
```

```{r mix04-water-pos-amp-ms2-mbank, echo = FALSE, fig.width = 5, fig.height = 5}
std_mbank <- get_mbank(mbank, inchikey = unique(std_hmdb$inchikey))
sim <- plot_ms2_similarity_heatmap(std_ms2, std_mbank, csp)
```

```{r mix04-water-pos-amp-mirror-mbank, echo = TRUE, fig.cap = "Mirror plots"}
tmp_sel <- plot_select_ms2(std_ms2, std_mbank, 0.7,
                           ppm = csp@ppm, tolerance = csp@tolerance)
```

In addition we match (**all**) the MS2 spectra for the matched features against
all spectra from HMDB or MassBank identifying reference spectra with a
similarity larger than 0.7. The results (if any spectra matched) are shown in
the two following tables.

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, hmdb, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, mbank, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

#### Summary

- FT05824 (RT=211.6, `[M+H]+` ion) with confidence **B**.
- FT06441 (RT=212.2, `[M+Na]+` ion) with confidence **B-**.
- FT06861 (RT=211.5, `[M+K]+` ion) with confidence **B-**.
- Eventually FT07015 (RT=226.4, `[M+2Na-H]+` ion) with confidence **B-**.
- Reference MS2: `[M+H]+`: FT05824_F07.S1006, FT05824_F07.S1049,
  FT05824_F08.S1004; high confidence. `[M+Na]+`: FT06441_F08.S0996; low
  confidence. `[M+K]+`: FT06861_F07.S1008, FT06861_F08.S1005; low confidence.

```{r, echo = FALSE}
## Define the feature (ion) and MS2 spectra to insert
fts <- data.frame(feature_id = c("FT05824", "FT06441", "FT06861"),
                  confidence_level = c("B", "B-", "B-"))

## Get MS2 spectra:
ms2 <- std_ms2[c("FT05824_F07.S1006", "FT05824_F07.S1049", "FT05824_F08.S1004",
                 "FT06441_F08.S0996", "FT06861_F07.S1008", "FT06861_F08.S1005")]
ms2$confidence <- c(rep("high", 3), rep("low", 3))
```

```{r, add-ions, echo = FALSE}
```

```{r, add-ms2-spectra, echo = FALSE}
```


### Glutamine

```{r, echo = FALSE}
std <- "Glutamine"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

Many features have been assigned to this standard based on m/z matching.
Based on their retention times, these seem however to represent ions from
different compounds.

We next plot the EIC for the assigned feature and visually inspect these.

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WP", std_ms2)
```

The cleaned MS2 spectra for the features matched to `r std` are shown below
(peaks with an intensity below 5% of the maximum peak and spectra with less
than 2 peaks were removed).

```{r mix04-water-pos-glutamine-ms2, echo = FALSE}
plot_spectra(std_ms2)
```

We next match the extracted MS2 spectra against the reference spectra for 
`r std` from HMDB and MassBank.

```{r mix04-water-pos-glutamine-ms2-hmdb, echo = FALSE, fig.width = 5, fig.height = 5}
hmdb_id <- std_dilution$HMDB[std_dilution$name == std]
std_hmdb <- Spectra(cdb, filter = ~ compound_id == hmdb_id)
sim <- plot_ms2_similarity_heatmap(std_ms2, std_hmdb, csp)
```

```{r mix04-water-pos-glutamine-mirror-hmdb, echo = TRUE, fig.cap = "Mirror plots"}
std_ms2_sel <- plot_select_ms2(std_ms2, std_hmdb, 0.8, ppm = csp@ppm,
                               tolerance = csp@tolerance)
```

```{r mix04-water-pos-glutamine-ms2-mbank, echo = FALSE, fig.width = 5, fig.height = 5}
std_mbank <- get_mbank(mbank, inchikey = unique(std_hmdb$inchikey))
sim <- plot_ms2_similarity_heatmap(std_ms2, std_mbank, csp)
```

```{r mix04-water-pos-glutamine-mirror-mbank, echo = TRUE, fig.cap = "Mirror plots"}
tmp_sel <- plot_select_ms2(std_ms2, std_mbank, 0.7, ppm = csp@ppm,
                               tolerance = csp@tolerance)
```

In addition we match (**all**) the MS2 spectra for the matched features against
all spectra from HMDB or MassBank identifying reference spectra with a
similarity larger than 0.7. The results (if any spectra matched) are shown in
the two following tables.

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, hmdb, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, mbank, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```


#### Summary

- FT01239 (RT = 180.6 , ion `[M+H]+`): confidence **A** (A grading because the
  MS2 spectra for that ion do really only match those of the standard).
- FT00841 (RT = 180.4, ion `[M+H-NH3]+`): confidence **A-**.
- Reference MS2: `[M+H]+`: FT01239_F07.S0885, FT01239_F08.S0877; high
  confidence. `[M+H-NH3]+`: FT00841_F07.S0884, FT00841_F08.S0876; low
  confidence.

```{r, echo = FALSE}
## Define the feature (ion) and MS2 spectra to insert
fts <- data.frame(feature_id = c("FT01239", "FT00841"),
                  confidence_level = c("A", "A-"))

## Get MS2 spectra:
ms2 <- std_ms2[c("FT01239_F07.S0885", "FT01239_F08.S0877", "FT00841_F07.S0884",
                 "FT00841_F08.S0876")]
ms2$confidence <- c(rep("high", 2), rep("low", 2))
```

```{r, add-ions, echo = FALSE}
```

```{r, add-ms2-spectra, echo = FALSE}
```


### GMP

```{r, echo = FALSE}
std <- "GMP"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

Many features have been assigned to this standard based on
m/z matching. Based on their retention times, these seem however to represent
ions from two different compounds.

We next plot the EIC for the assigned feature and visually inspect these.

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WP", std_ms2)
```

The cleaned MS2 spectra for the features matched to `r std` are shown below
(peaks with an intensity below 5% of the maximum peak and spectra with less
than 2 peaks were removed).

```{r mix04-water-pos-gmp-ms2, echo = FALSE}
plot_spectra(std_ms2)
```

We next match the extracted MS2 spectra against the reference spectra for 
`r std` from HMDB and MassBank.

```{r mix04-water-pos-gmp-ms2-hmdb, echo = FALSE, fig.width = 5, fig.height = 5}
hmdb_id <- std_dilution$HMDB[std_dilution$name == std]
std_hmdb <- Spectra(cdb, filter = ~ compound_id == hmdb_id)
sim <- plot_ms2_similarity_heatmap(std_ms2, std_hmdb, csp)
```

We find matches for spectra from feature FT06285 and FT06288. The best of them
are shown below.

```{r mix04-water-pos-gmp-mirror-hmdb, echo = TRUE, fig.cap = "Mirror plots"}
std_ms2_sel <- plot_select_ms2(std_ms2, std_hmdb, 0.8, ppm = csp@ppm,
                               tolerance = csp@tolerance)
```

```{r mix04-water-pos-gmp-ms2-mbank, echo = FALSE, fig.width = 5, fig.height = 5}
std_mbank <- get_mbank(mbank, inchikey = unique(std_hmdb$inchikey))
sim <- plot_ms2_similarity_heatmap(std_ms2, std_mbank, csp)
```

We find similar results for MassBank.

In addition we match (**all**) the MS2 spectra for the matched features against
all spectra from HMDB or MassBank identifying reference spectra with a
similarity larger than 0.7. The results (if any spectra matched) are shown in
the two following tables.

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, hmdb, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, mbank, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```


#### Summary

- FT06285 (RT=225.8, `[M+H]+` ion) confidence level **A**.
- FT07248 (RT=225.9, `[M+K]+` ion) confidence level **A-**.
- FT06864 (RT=225.9, `[M+Na]+` ion) confidence level **A-**.
- Reference MS2: `[M+H]+`: FT06285_F07.S1055, FT06285_F07.S1111,
  FT06285_F08.S1047, FT06285_F08.S1152; high confidence. 

```{r, echo = FALSE}
## Define the feature (ion) and MS2 spectra to insert
fts <- data.frame(feature_id = c("FT06285", "FT07248", "FT06864"),
                  confidence_level = c("A", "A-", "A-"))

## Get MS2 spectra:
ms2 <- std_ms2[c("FT06285_F07.S1055", "FT06285_F07.S1111", "FT06285_F08.S1047",
                 "FT06285_F08.S1152")]
ms2$confidence <- c(rep("high", 4))
```

```{r, add-ions, echo = FALSE}
```

```{r, add-ms2-spectra, echo = FALSE}
```

  
### Leucine

```{r, echo = FALSE}
std <- "Leucine"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

Many features have been assigned to this standard based on
m/z matching. Based on their retention times, these seem however to represent
ions from different compounds.

We next plot the EIC for the assigned feature and visually inspect these.

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WP", std_ms2)
```

The cleaned MS2 spectra for the features matched to `r std` are shown below
(peaks with an intensity below 5% of the maximum peak and spectra with less
than 2 peaks were removed).

```{r mix04-water-pos-leucine-ms2, echo = FALSE}
plot_spectra(std_ms2)
```

We next match the extracted MS2 spectra against the reference spectra for 
`r std` from HMDB and MassBank.

```{r mix04-water-pos-leucine-ms2-hmdb, echo = FALSE, fig.width = 5, fig.height = 5}
hmdb_id <- std_dilution$HMDB[std_dilution$name == std]
std_hmdb <- Spectra(cdb, filter = ~ compound_id == hmdb_id)
sim <- plot_ms2_similarity_heatmap(std_ms2, std_hmdb, csp)
```

The best matching spectrum is F07.S0659 and F08.S0644 from FT0790.
Also F08.S0257 and F07.S0255 from FT0789 match (with similarity around 0.65).
 
```{r mix04-water-pos-leucine-mirror-hmdb, echo = TRUE, fig.cap = "Mirror plots"}
std_ms2_sel <- plot_select_ms2(std_ms2, std_hmdb, 0.7,
                               ppm = csp@ppm, tolerance = csp@tolerance)
```

```{r mix04-water-pos-leucine-acid-ms2-mbank, echo = FALSE, fig.width = 5, fig.height = 5}
std_mbank <- get_mbank(mbank, inchikey = unique(std_hmdb$inchikey))
sim <- plot_ms2_similarity_heatmap(std_ms2, std_mbank, csp)
```

In addition we match (**all**) the MS2 spectra for the matched features against
all spectra from HMDB or MassBank identifying reference spectra with a
similarity larger than 0.7. The results (if any spectra matched) are shown in
the two following tables.

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, hmdb, sv = c("rtime", "target_name", "target_inchikey", "score"),
              name = "target_name", param = csp)
```

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, mbank,
              sv = c("rtime", "target_name", "target_inchikey", "score"),
              name = "target_name", param = csp)
```

There are matches with similarity higher than 0.7 involving more than one
compound (maybe they are similar?).


#### Summary

- FT00903 (RT = 148.1, `[M+H]+`): confidence level **B**.
- FT01380 (RT = 149.3, `[M+Na]+`): confidence level **B-**.
- FT01825 (RT = 149.3, `[M+2Na-H]+`): confidence level **B-**.
- FT02414 (RT = 149.3, `[M+2K-H]+`): confidence level **B-**.
- Reference MS2: `[M+H]+`: FT00903_F08.S0677; high confidence. `[M+2Na-H]+`:
  FT01825_F08.S0713 low confidence.

```{r, echo = FALSE}
## Define the feature (ion) and MS2 spectra to insert
fts <- data.frame(feature_id = c("FT00903", "FT01380", "FT01825", "FT02414"),
                  confidence_level = c("B", "B-", "B-", "B-"))

## Get MS2 spectra:
ms2 <- std_ms2[c("FT00903_F08.S0677", "FT01825_F08.S0713")]
ms2$confidence <- c("high", "low")
```

```{r, add-ions, echo = FALSE}
```

```{r, add-ms2-spectra, echo = FALSE}
```


### Phenylethylamine

```{r, echo = FALSE}
std <- "Phenylethylamine"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

Many features have been assigned to this standard based on
m/z matching. Based on their retention times, these seem however to represent
ions from different compounds.

We next plot the EIC for the assigned feature and visually inspect these.

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WP", std_ms2)
```

The cleaned MS2 spectra for the features matched to `r std` are shown below
(peaks with an intensity below 5% of the maximum peak and spectra with less
than 2 peaks were removed).

```{r mix04-water-pos-phenylethylamine-ms2, echo = FALSE}
plot_spectra(std_ms2)
```

We next match the extracted MS2 spectra against the reference spectra for 
`r std` from HMDB and MassBank.

```{r mix04-water-pos-phenylethylamine-ms2-hmdb, echo = FALSE, fig.width = 5, fig.height = 5}
hmdb_id <- std_dilution$HMDB[std_dilution$name == std]
std_hmdb <- Spectra(cdb, filter = ~ compound_id == hmdb_id)
sim <- plot_ms2_similarity_heatmap(std_ms2, std_hmdb, csp)
```

All extracted spectra seem to match reference spectra from `r std` (although
all of the associated features were matched to ions different from `[M+H]+`.  
 
```{r mix04-water-pos-phenylethylamine-hmdb, echo = TRUE, fig.cap = "Mirror plots"}
std_ms2_sel <- plot_select_ms2(std_ms2, std_hmdb, 0.8, ppm = csp@ppm,
                               tolerance = csp@tolerance)
```

```{r mix04-water-pos-phenylethylamine-ms2-mbank, echo = FALSE, fig.width = 5, fig.height = 5}
std_mbank <- get_mbank(mbank, inchikey = unique(std_hmdb$inchikey))
sim <- plot_ms2_similarity_heatmap(std_ms2, std_mbank, csp)
```

In addition we match (**all**) the MS2 spectra for the matched features against
all spectra from HMDB or MassBank identifying reference spectra with a
similarity larger than 0.7. The results (if any spectra matched) are shown in
the two following tables.

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, hmdb, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, mbank, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```


#### Summary

- Inconclusive: from the EIC it looks the peak is around 37 seconds but then
  (column overload?) signal goes until 126 seconds. All MS2 match nicely, but
  the difference in precursor m/z is ~ 17.03 (against the **assumed** `[M+H]+`
  ion of Phenylethylamine). This difference matches exactly the mass of
  `NH3`. Thus, most likely, the MS2 spectra from HMDB do in fact represent also
  `[M+H-NH3]+` ions which would make the MS2 spectra matching perfectly.
- Tricky situation is that none of the features' RT would actually match the
  correct one.
- FT00449 (RT = 46.62, ion `[M+H-NH3]+`): confidence level **B**?
- FT00448 (RT = 122.6, ion `[M+H-NH3]+`): confidence level **B**?
- FT00450 (RT = 73.43, ion `[M+H-NH3]+`): confidence level **B**?
- Reference MS2: `[M+H-NH3]+`: FT00449_F07.S0121, FT00449_F07.S0213,
  FT00448_F07.S0547; high confidence).

```{r, echo = FALSE}
## Get MS2 spectra:
ms2 <- std_ms2[c("FT00449_F07.S0121", "FT00449_F07.S0213", "FT00448_F07.S0547")]
ms2$confidence <- c("high")
```

```{r, add-ms2-spectra, echo = FALSE}
```


### Putrescine

```{r, echo = FALSE}
std <- "Putrescine"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

Many features have been assigned to this standard based on m/z matching.
Based on their retention times, these seem however to represent ions from
different compounds.

We next plot the EIC for the assigned feature and visually inspect these.

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WP", std_ms2)
```

The cleaned MS2 spectra for the features matched to `r std` are shown below
(peaks with an intensity below 5% of the maximum peak and spectra with less
than 2 peaks were removed).

```{r mix04-water-pos-putrescine-ms2, echo = FALSE}
plot_spectra(std_ms2)
```


#### Summary

- FT00245 (RT = 175.6, `[M+H]+` ion): confidence level **D**.
- FT00093 (RT = 174.2, `[M+H-NH3]+` ion): confidence level **D**.

```{r, echo = FALSE}
## Define the feature (ion) and MS2 spectra to insert
fts <- data.frame(feature_id = c("FT00245", "FT00093"),
                  confidence_level = c("D", "D"))
```

```{r, add-ions, echo = FALSE}
```


### Pyridoxine

```{r, echo = FALSE}
std <- "Pyridoxine"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

Many features have been assigned to this standard based on m/z matching.
Based on their retention times, these seem however to represent ions from
different compounds.

We next plot the EIC for the assigned feature and visually inspect these.

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WP", std_ms2)
```

The cleaned MS2 spectra for the features matched to `r std` are shown below
(peaks with an intensity below 5% of the maximum peak and spectra with less
than 2 peaks were removed).

```{r mix04-water-pos-pyridoxine-ms2, echo = FALSE}
plot_spectra(std_ms2)
```

We next match the extracted MS2 spectra against the reference spectra for 
`r std` from HMDB and MassBank.

```{r mix04-water-pos-pyridoxine-ms2-hmdb, echo = FALSE, fig.width = 5, fig.height = 5}
hmdb_id <- std_dilution$HMDB[std_dilution$name == std]
std_hmdb <- Spectra(cdb, filter = ~ compound_id == hmdb_id)
sim <- plot_ms2_similarity_heatmap(std_ms2, std_hmdb, csp)
```

```{r mix04-water-pos-pyridoxine-hmdb, echo = TRUE, fig.cap = "Mirror plots"}
std_ms2_sel <- plot_select_ms2(std_ms2, std_hmdb, 0.85, ppm = csp@ppm,
                               tolerance = csp@tolerance)
```

```{r mix04-water-pos-pyridoxine-ms2-mbank, echo = FALSE, fig.width = 5, fig.height = 5}
std_mbank <- get_mbank(mbank, inchikey = unique(std_hmdb$inchikey))
sim <- plot_ms2_similarity_heatmap(std_ms2, std_mbank, csp)
```

In addition we match (**all**) the MS2 spectra for the matched features against
all spectra from HMDB or MassBank identifying reference spectra with a
similarity larger than 0.7. The results (if any spectra matched) are shown in
the two following tables.

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, hmdb, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, mbank, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

#### Summary

- FT01691 (RT = 108.5, `[M+H]+` ion) with confidence level **A**.
- FT01347 (RT = 110.8, `[M+H-H2O]+` ion) with confidence level **A**.
- Reference MS2: `[M+H]+`: FT01691_F07.S0447, FT01691_F07.S0484,
  FT01691_F07.S0512, FT01691_F08.S0426, FT01691_F08.S0463, FT01691_F08.S0494;
  high confidence. `[M+H-H2O]+`: FT01347_F07.S0458, FT01347_F07.S0507,
  FT01347_F08.S0434; high confidence.

```{r, echo = FALSE}
## Define the feature (ion) and MS2 spectra to insert
fts <- data.frame(feature_id = c("FT01691", "FT01347"),
                  confidence_level = c("A", "A"))

## Get MS2 spectra:
ms2 <- std_ms2[c("FT01691_F07.S0447", "FT01691_F07.S0484", "FT01691_F07.S0512",
                 "FT01691_F08.S0426", "FT01691_F08.S0463", "FT01691_F08.S0494",
                 "FT01347_F07.S0458", "FT01347_F07.S0507", "FT01347_F08.S0434")]
ms2$confidence <- c("high")
```

```{r, add-ions, echo = FALSE}
```

```{r, add-ms2-spectra, echo = FALSE}
```


### Riboflavin

```{r, echo = FALSE}
std <- "Riboflavin"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

Many features have been assigned to this standard based on
m/z matching. Based on their retention times, these seem however to represent
ions from different compounds.

We next plot the EIC for the assigned feature and visually inspect these.

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WP", std_ms2)
```

The cleaned MS2 spectra for the features matched to `r std` are shown below
(peaks with an intensity below 5% of the maximum peak and spectra with less
than 2 peaks were removed).

```{r mix04-water-pos-riboflavin-ms2, echo = FALSE}
plot_spectra(std_ms2)
```

We next match the extracted MS2 spectra against the reference spectra for 
`r std` from HMDB and MassBank.

```{r mix04-water-pos-riboflavin-ms2-hmdb, echo = FALSE, fig.width = 5, fig.height = 5}
hmdb_id <- std_dilution$HMDB[std_dilution$name == std]
std_hmdb <- Spectra(cdb, filter = ~ compound_id == hmdb_id)
sim <- plot_ms2_similarity_heatmap(std_ms2, std_hmdb, csp)
```

```{r mix04-water-pos-riboflavin-hmdb, echo = TRUE, fig.cap = "Mirror plots"}
std_ms2_sel <- plot_select_ms2(std_ms2, std_hmdb, 0.7, ppm = csp@ppm,
                               tolerance = csp@tolerance)
```

We have very good matches for spectra from FT06639 and lower similarity ones
for those from FT06141.

```{r mix04-water-pos-riboflavin-ms2-mbank, echo = FALSE, fig.width = 5, fig.height = 5}
std_mbank <- get_mbank(mbank, inchikey = unique(std_hmdb$inchikey))
sim <- plot_ms2_similarity_heatmap(std_ms2, std_mbank, csp)
```

In addition we match (**all**) the MS2 spectra for the matched features against
all spectra from HMDB or MassBank identifying reference spectra with a
similarity larger than 0.7. The results (if any spectra matched) are shown in
the two following tables.

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, hmdb, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, mbank, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```


#### Summary

- FT06639 (RT=145.7, `[M+H]+` ion) with confidence level **A**.
- FT07168 (RT=144.7, `[M+Na]+` ion) with confidence level **A-**.
- FT07654 (RT=145.7, `[M+2Na-H]+` ion) with confidence level **A-**.
- Reference MS2: `[M+H]+`: FT06639_F07.S0686, FT06639_F08.S0681; high
  confidence. `[M+Na]+`: FT07168_F07.S0689, FT07168_F08.S0684; low confidence.
  `[M+2Na-H]+`: FT07654_F07.S0691, FT07654_F08.S0686; low confidence.

```{r, echo = FALSE}
## Define the feature (ion) and MS2 spectra to insert
fts <- data.frame(feature_id = c("FT07654"),
                  confidence_level = c("A-"))

## Get MS2 spectra:
ms2 <- std_ms2[c("FT06639_F07.S0686", "FT06639_F08.S0681", "FT07168_F07.S0689",
                 "FT07168_F08.S0684", "FT07654_F07.S0691", "FT07654_F08.S0686")]
ms2$confidence <- c(rep("high", 2), rep("low", 4))
```

```{r, add-ions, echo = FALSE}
```

```{r, add-ms2-spectra, echo = FALSE}
```


### Ribose

```{r, echo = FALSE}
std <- "Ribose"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

We next plot the EIC for the assigned features and visually inspect these.

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WP", std_ms2)
```

The cleaned MS2 spectra for the features matched to `r std` are shown below
(peaks with an intensity below 5% of the maximum peak and spectra with less
than 2 peaks were removed).

```{r mix04-water-pos-ribose-ms2, echo = FALSE}
plot_spectra(std_ms2)
```

#### Summary

- Inconclusive. No MS2 spectra available and no clear, unambiguous, EIC signal.


### SAMe

```{r, echo = FALSE}
std <- "SAMe"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

Many features have been assigned to this standard based on m/z matching. Based
on their retention times, these seem however to represent ions from different
compounds.

We next plot the EIC for the assigned feature and visually inspect these.

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WP", std_ms2)
```

The only available MS2 spectrum (cleaned) for the features matched to `r std` is
shown below (peaks with an intensity below 5% of the maximum peak and spectra
with less than 2 peaks were removed).

```{r mix04-water-pos-same-ms2, echo = FALSE}
plot_spectra(std_ms2)
```

We next match the extracted MS2 spectra against the reference spectra for 
`r std` from HMDB and MassBank.

```{r mix04-water-pos-same-ms2-hmdb, echo = FALSE, fig.width = 5, fig.height = 5}
hmdb_id <- std_dilution$HMDB[std_dilution$name == std]
std_hmdb <- Spectra(cdb, filter = ~ compound_id == hmdb_id)
sim <- plot_ms2_similarity_heatmap(std_ms2, std_hmdb, csp)
```

```{r mix04-water-pos-same-ms2-mbank, echo = FALSE, fig.width = 5, fig.height = 5}
std_mbank <- get_mbank(mbank, inchikey = unique(std_hmdb$inchikey))
sim <- plot_ms2_similarity_heatmap(std_ms2, std_mbank, csp)
```

In both cases we don't find any match. In addition, since the extracted
spectrum is associated to an ion different from `[M+H]+` we perform a neutral
loss spectra comparison.

```{r mix04-water-pos-same-ms2-hmdb_nl, echo = FALSE, fig.width = 5, fig.height = 5}
std_hmdb_nl <- hmdb_nl[hmdb_nl$compound_id == hmdb_id]
std_ms2_nl <- neutralLoss(std_ms2, nl_param)
sim <- plot_ms2_similarity_heatmap(std_ms2_nl, std_hmdb_nl, csp_nl)
```

```{r mix04-water-pos-same-ms2-mbank_nl, echo = FALSE, fig.width = 5, fig.height = 5}
std_mbank_nl <- get_mbank(mbank, inchikey = unique(std_hmdb$inchikey), nl = TRUE)
std_ms2_nl <- neutralLoss(std_ms2, nl_param)
sim <- plot_ms2_similarity_heatmap(std_ms2_nl, std_mbank_nl, csp_nl)
```

We only find very low similarity matches with the HMDB comparison.

In addition we match (**all**) the MS2 spectra for the matched features against
all spectra from HMDB or MassBank identifying reference spectra with a
similarity larger than 0.7. The results (if any spectra matched) are shown in
the two following tables.

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, hmdb, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, mbank, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```


#### Summary

- Inconclusive. No MS2 spectra available for the most likely ions (see
  below). Need to cross-check with negative polarity and serum samples.
- FT07201 (RT = 197.3, `[M+H]+` ion): confidence level **D**.
- FT02280 (RT = 197.3, `[M+2H]2+` ion): confidence level **D**.

```{r, echo = FALSE}
## Define the feature (ion) and MS2 spectra to insert
fts <- data.frame(feature_id = c("FT07201", "FT02280"),
                  confidence_level = c("D", "D"))
```

```{r, add-ions, echo = FALSE}
```


### Sedoheptulose-7-phosphate

```{r, echo = FALSE}
std <- "Sedoheptulose-7-phosphate"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

Many features have been assigned to this standard based on m/z matching.
Based on their retention times, these seem however to represent ions from
different compounds.

We next plot the EIC for the assigned feature and visually inspect these.

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WP", std_ms2)
```

The cleaned MS2 spectra for the features matched to `r std` are shown below
(peaks with an intensity below 5% of the maximum peak and spectra with less
than 2 peaks were removed).

```{r mix04-water-pos-sedoheptulose-7-phosphate-ms2, echo = FALSE}
plot_spectra(std_ms2)
```

We next match the extracted MS2 spectra against the reference spectra for 
`r std` from HMDB and MassBank.

```{r mix04-water-pos-sedoheptulose-7-phosphate-ms2-hmdb, echo = FALSE, fig.width = 5, fig.height = 5}
hmdb_id <- std_dilution$HMDB[std_dilution$name == std]
std_hmdb <- Spectra(cdb, filter = ~ compound_id == hmdb_id)
sim <- plot_ms2_similarity_heatmap(std_ms2, std_hmdb, csp)
```

```{r mix04-water-pos-sedoheptulose-7-phosphate-ms2-mbank, echo = FALSE, fig.width = 5, fig.height = 5}
std_mbank <- get_mbank(mbank, inchikey = unique(std_hmdb$inchikey))
sim <- plot_ms2_similarity_heatmap(std_ms2, std_mbank, csp)
```

It seems like there are no reference spectra for `r std` both in HMDB
and MassBank?

In addition we match (**all**) the MS2 spectra for the matched features against
all spectra from HMDB or MassBank identifying reference spectra with a
similarity larger than 0.7. The results (if any spectra matched) are shown in
the two following tables.

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, hmdb, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, mbank, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

#### Summary

- Needs to be cross-checked with negative polarity.
- FT03777 (RT = 231.6, `[M+H-H2O]+` ion): confidence level **D**.
- FT04240 (RT = 231.1, `[M+H]+` ion): confidence level **D**.
- FT04851 (RT = 231.9, `[M+Na]+` ion): confidence level **D**.
- FT05294 (RT = 232, `[M+K]+` ion): confidence level **D**.
- Reference MS2: `[M+H-H2O]+`: FT03777_F07.S1079, FT03777_F08.S1083; low
  confidence. `[M+Na]+`: FT04851_F07.S1080; low confidence. `[M+K]+`:
  FT05294_F08.S1111; low confidence.

```{r, echo = FALSE}
## Define the feature (ion) and MS2 spectra to insert
fts <- data.frame(feature_id = c("FT03777", "FT04240", "FT04851", "FT05294"),
                  confidence_level = c("D", "D", "D", "D"))

## Get MS2 spectra:
ms2 <- std_ms2[c("FT03777_F07.S1079", "FT03777_F08.S1083", "FT04851_F07.S1080",
                 "FT05294_F08.S1111")]
ms2$confidence <- c(rep("low", 4))
```

```{r, add-ions, echo = FALSE}
```

```{r, add-ms2-spectra, echo = FALSE}
```


# Water, negative polarity

We now perform the analysis on the samples with the standards still solved in
pure water but acquired in negative polarity mode.

```{r}
POLARITY <- "NEG"
data <- filterFile(data_all, which(data_all$polarity == POLARITY))
hmdb <- hmdb_neg
hmdb_nl <- hmdb_neg_nl
```

```{r, preprocessing, eval = !file.exists(paste0(RDATA_PATH, "processed_data_", POLARITY, ".RData")), message = FALSE}
```

```{r, echo = FALSE}
load(paste0(RDATA_PATH, paste0("processed_data_", POLARITY, ".RData")))
data_FS <- filterFile(data, which(data$mode == "FS"),
                      keepFeatures = TRUE)
```


## Signal intensity difference

We compute the difference in (log2) signals between samples with high and low
concentration of the standards and calculate the p-value for this difference
using the Student's t-test.

```{r, abundance-difference}
```

## Identification of features matching standards

We now identify all features matching the any of the pre-defined set of adducts
for the standards of mix `r MIX`. Matching features are further subsetted to
those that show an at least twice as high signal in samples with high
concentration compared to those with low concentrations. Also, features with a
signal in low concentration but no detectable signal in high concentration are
removed.

```{r, define-adducts-neg}
```
```{r, match-features}
```

For most of the standards (`r length(unique(mD$target_name))` out of 
`r nrow(std_dilution)`) in mix `r MIX` at least one feature was found
matching the standards adducts' m/z. 

```{r, table-standard-no-feature, results = "asis", echo = FALSE}
```

In the next sections we investigate for each standard which assignment would be
the correct one or, for those for which no signal was detected, why that was the
case.

## Standards with matching features

### Acetylalanine

```{r, echo = FALSE}
std <- "Acetylalanine"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

Many features have been assigned to this standard based on
m/z matching. Based on their retention times, these seem however to represent
ions from different compounds.

We next plot the EICs for the assigned features and visually inspect these.

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WN", std_ms2)
```

The cleaned MS2 spectra for the features matched to `r std` are shown below
(peaks with an intensity below 5% of the maximum peak and spectra with less
than 2 peaks were removed).

```{r mix04-water-neg-acetylalanine-ms2, echo = FALSE}
plot_spectra(std_ms2)
```

We next match the extracted MS2 spectra against the reference spectra for 
`r std` from HMDB and MassBank.

```{r mix04-water-neg-acetylalanine-ms2-hmdb, echo = FALSE, fig.width = 5, fig.height = 5}
hmdb_id <- std_dilution$HMDB[std_dilution$name == std]
std_hmdb <- Spectra(cdb, filter = ~ compound_id == hmdb_id)
sim <- plot_ms2_similarity_heatmap(std_ms2, std_hmdb, csp)
```

MS2 spectra for features FT0176 and FT0175 match the reference spectra well.

```{r mix04-water-neg-acetylalanine-mirror-hmdb, echo = TRUE, fig.cap = "Mirror plots"}
std_ms2_sel <- plot_select_ms2(std_ms2, std_hmdb, 0.7,
                               ppm = csp@ppm, tolerance = csp@tolerance)
```

```{r mix04-water-neg-acetylalanine-acid-ms2-mbank, echo = FALSE, fig.width = 5, fig.height = 5}
std_mbank <- get_mbank(mbank, inchikey = unique(std_hmdb$inchikey))
sim <- plot_ms2_similarity_heatmap(std_ms2, std_mbank, csp)
```

No spectrum matches reference spectra from MassBank.

In addition we match (**all**) the MS2 spectra for the matched features against
all spectra from HMDB or MassBank identifying reference spectra with a
similarity larger than 0.7. The results (if any spectra matched) are shown in
the two following tables.

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, hmdb, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, mbank, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```


#### Summary

- FT0175 (RT = 37.2, `[M-H]-` ion) with confidence level **B**.
- FT0604 (RT = 37, `[M+HCOO]-` ion) with confidence level **B-**.
- Reference MS2: `[M-H]-`: FT0175_F15.S0104, FT0175_F15.S0152; high confidence.

```{r, echo = FALSE}
## Define the feature (ion) and MS2 spectra to insert
fts <- data.frame(feature_id = c("FT0175", "FT0604"),
                  confidence_level = c("B", "B-"))

## Get MS2 spectra:
ms2 <- std_ms2[c("FT0175_F15.S0104", "FT0175_F15.S0152")]
ms2$confidence <- c("high")
```

```{r, add-ions, echo = FALSE}
```

```{r, add-ms2-spectra, echo = FALSE}
```


### Allantoin

```{r, echo = FALSE}
std <- "Allantoin"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WN", std_ms2)
```

The cleaned MS2 spectra for the features matched to `r std` are shown below
(peaks with an intensity below 5% of the maximum peak and spectra with less
than 2 peaks were removed).

```{r mix04-water-neg-allantoin-ms2, echo = FALSE}
plot_spectra(std_ms2)
```

We next match the extracted MS2 spectra against the reference spectra for 
`r std` from HMDB and MassBank.

```{r mix04-water-neg-gamma-aminobutyric-acid-ms2-hmdb, echo = FALSE, fig.width = 5, fig.height = 5}
hmdb_id <- std_dilution$HMDB[std_dilution$name == std]
std_hmdb <- Spectra(cdb, filter = ~ compound_id == hmdb_id)
sim <- plot_ms2_similarity_heatmap(std_ms2, std_hmdb, csp)
```

The two extracted spectra match HMDB reference spectra from `r std` but with
low similarity score. The best matches are shown below.

```{r mix04-water-neg-allantoin-hmdb, echo = TRUE, fig.cap = "Mirror plots"}
std_ms2_sel <- plot_select_ms2(std_ms2, std_hmdb, 0.4, ppm = csp@ppm,
                               tolerance = csp@tolerance)
```

The spectra seem to nicely match although the score is low.

```{r mix04-water-neg-allantoin-ms2-mbank, echo = FALSE, fig.width = 5, fig.height = 5}
std_mbank <- get_mbank(mbank, inchikey = unique(std_hmdb$inchikey))
sim <- plot_ms2_similarity_heatmap(std_ms2, std_mbank, csp)
```

We don't find any match with the MassBank comparison.

In addition we match (**all**) the MS2 spectra for the matched features against
all spectra from HMDB or MassBank identifying reference spectra with a
similarity larger than 0.7. The results (if any spectra matched) are shown in
the two following tables.

```{r, echo = FALSE, message = FALSE, results = "asis"}
std_ms2_sel <- Spectra()
perform_match(std_ms2, hmdb, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, mbank, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

#### Summary

- FT0420 (RT = 139.1, `[M-H]-` ion) with confidence level **C**.
- FT0760 (RT = 138.8, `[M+Cl]-` ion) with confidence level **C-**.
- FT0885 (RT = 138.8, `[M+HCOO]-` ion) with confidence level **C-**.
- Reference MS2: `[M-H-]`: FT0420_F15.S0420, FT0420_F16.S0418; low confidence.

```{r, echo = FALSE}
## Define the feature (ion) and MS2 spectra to insert
fts <- data.frame(feature_id = c("FT0420", "FT0760", "FT0885"),
                  confidence_level = c("C", "C-", "C-"))

## Get MS2 spectra:
ms2 <- std_ms2[c("FT0420_F15.S0420", "FT0420_F16.S0418")]
ms2$confidence <- c("low")
```

```{r, add-ions, echo = FALSE}
```

```{r, add-ms2-spectra, echo = FALSE}
```


### Gamma-Aminobutyric acid

```{r, echo = FALSE}
std <- "Gamma-Aminobutyric acid"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WN", std_ms2)
```

The cleaned MS2 spectra for the features matched to `r std` are shown below
(peaks with an intensity below 5% of the maximum peak and spectra with less
than 2 peaks were removed).

```{r mix04-water-neg-gamma-aminobutyric-acid-ms2, echo = FALSE}
plot_spectra(std_ms2)
```

#### Summary

- Inconclusive: no MS2 available and EICs are not clear.


### AMP

```{r, echo = FALSE}
std <- "AMP"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

One single feature has been assigned to this standard based on m/z matching.
Based on their retention times, these seem however to represent ions from
different compounds.

We next plot the EIC for the assigned feature and visually inspect these.

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WN", std_ms2)
```

The cleaned MS2 spectra for the features matched to `r std` are shown below
(peaks with an intensity below 5% of the maximum peak and spectra with less
than 2 peaks were removed).

```{r mix04-water-neg-amp-ms2, echo = FALSE}
plot_spectra(std_ms2)
```

We next match the extracted MS2 spectra against the reference spectra for 
`r std` from HMDB and MassBank.

```{r mix04-water-neg-amp-ms2-hmdb, echo = FALSE, fig.width = 5, fig.height = 5}
hmdb_id <- std_dilution$HMDB[std_dilution$name == std]
std_hmdb <- Spectra(cdb, filter = ~ compound_id == hmdb_id)
sim <- plot_ms2_similarity_heatmap(std_ms2, std_hmdb, csp)
```

All the extracted spectra seem to match HMDB reference spectra for `r std`.

```{r mix04-water-neg-amp-mirror-hmdb, echo = TRUE, fig.cap = "Mirror plots"}
std_ms2_sel <- plot_select_ms2(std_ms2, std_hmdb, 0.7,
                               ppm = csp@ppm, tolerance = csp@tolerance)
```

We find similar results for MassBank as shown below.

```{r mix04-water-neg-amp-ms2-mbank, echo = FALSE, fig.width = 5, fig.height = 5}
std_mbank <- get_mbank(mbank, inchikey = unique(std_hmdb$inchikey))
sim <- plot_ms2_similarity_heatmap(std_ms2, std_mbank, csp)
```

```{r mix04-water-neg-amp-mirror-mbank, echo = TRUE, fig.cap = "Mirror plots"}
tmp_sel <- plot_select_ms2(std_ms2, std_mbank, 0.7,
                           ppm = csp@ppm, tolerance = csp@tolerance)
```

In addition we match (**all**) the MS2 spectra for the matched features against
all spectra from HMDB or MassBank identifying reference spectra with a
similarity larger than 0.7. The results (if any spectra matched) are shown in
the two following tables.

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, hmdb, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, mbank, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

#### Summary

- FT2848 (RT = 211.3, `[M-H]-` ion) with confidence level **B**.
- FT3563 (RT = 211.1, `[M+HCOO]-` ion) with confidence level **B-**.
- Reference MS2: `[M-H-]`: FT2848_F15.S0706, FT2848_F16.S0696 ; high confidence.

```{r, echo = FALSE}
## Define the feature (ion) and MS2 spectra to insert
fts <- data.frame(feature_id = c("FT2848", "FT3563"),
                  confidence_level = c("B", "B-"))

## Get MS2 spectra:
ms2 <- std_ms2[c("FT2848_F15.S0706", "FT2848_F16.S0696")]
ms2$confidence <- c("high")
```

```{r, add-ions, echo = FALSE}
```

```{r, add-ms2-spectra, echo = FALSE}
```


### Glutamine

```{r, echo = FALSE}
std <- "Glutamine"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

Many features have been assigned to this standard based on
m/z matching. Based on their retention times, these seem however to represent
ions from different compounds.

We next plot the EIC for the assigned feature and visually inspect these.

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WN", std_ms2)
```

The cleaned MS2 spectra for the features matched to `r std` are shown below
(peaks with an intensity below 5% of the maximum peak and spectra with less
than 2 peaks were removed).

```{r mix04-water-neg-glutamine-ms2, echo = FALSE}
plot_spectra(std_ms2)
```

We next match the extracted MS2 spectra against the reference spectra for 
`r std` from HMDB and MassBank.

```{r mix04-water-neg-glutamine-ms2-hmdb, echo = FALSE, fig.width = 5, fig.height = 5}
hmdb_id <- std_dilution$HMDB[std_dilution$name == std]
std_hmdb <- Spectra(cdb, filter = ~ compound_id == hmdb_id)
sim <- plot_ms2_similarity_heatmap(std_ms2, std_hmdb, csp)
```

```{r mix04-water-neg-glutamine-mirror-hmdb, echo = TRUE, fig.cap = "Mirror plots"}
std_ms2_sel <- plot_select_ms2(std_ms2, std_hmdb, 0.7, ppm = csp@ppm,
                               tolerance = csp@tolerance)
```

```{r mix04-water-neg-glutamine-ms2-mbank, echo = FALSE, fig.width = 5, fig.height = 5}
std_mbank <- get_mbank(mbank, inchikey = unique(std_hmdb$inchikey))
sim <- plot_ms2_similarity_heatmap(std_ms2, std_mbank, csp)
```

```{r mix04-water-neg-glutamine-mirror-mbank, echo = TRUE, fig.cap = "Mirror plots"}
tmp_sel <- plot_select_ms2(std_ms2, std_mbank, 0.4, ppm = csp@ppm,
                               tolerance = csp@tolerance)
```

In addition we match (**all**) the MS2 spectra for the matched features against
all spectra from HMDB or MassBank identifying reference spectra with a
similarity larger than 0.7. The results (if any spectra matched) are shown in
the two following tables.

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, hmdb, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, mbank, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

#### Summary

- FT0325 (RT = 180.4, `[M-H]-` ion) confidence level **B**.
- FT0651 (RT = 179.6 `[M+Cl]-` ion) confidence level **B-**.
- FT0751 (RT = 179.6 `[M+HCOO]-` ion) confidence level **B-**.
- Reference MS2: `[M-H]-`: FT0325_F15.S0601, FT0325_F16.S0591; high confidence.

```{r, echo = FALSE}
## Define the feature (ion) and MS2 spectra to insert
fts <- data.frame(feature_id = c("FT0325", "FT0651", "FT0751"),
                  confidence_level = c("B", "B-", "B-"))

## Get MS2 spectra:
ms2 <- std_ms2[c("FT0325_F15.S0601", "FT0325_F16.S0591")]
ms2$confidence <- c("high")
```

```{r, add-ions, echo = FALSE}
```

```{r, add-ms2-spectra, echo = FALSE}
```


### GMP

```{r, echo = FALSE}
std <- "GMP"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

Many features have been assigned to this standard based on m/z matching.
Based on their retention times, these seem however to represent ions from
different compounds.

We next plot the EIC for the assigned feature and visually inspect these.

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WN", std_ms2)
```

The cleaned MS2 spectra for the features matched to `r std` are shown below
(peaks with an intensity below 5% of the maximum peak and spectra with less
than 2 peaks were removed).

```{r mix04-water-neg-gmp-ms2, echo = FALSE}
plot_spectra(std_ms2)
```

We next match the extracted MS2 spectra against the reference spectra for 
`r std` from HMDB and MassBank.

```{r mix04-water-neg-gmp-ms2-hmdb, echo = FALSE, fig.width = 5, fig.height = 5}
hmdb_id <- std_dilution$HMDB[std_dilution$name == std]
std_hmdb <- Spectra(cdb, filter = ~ compound_id == hmdb_id)
sim <- plot_ms2_similarity_heatmap(std_ms2, std_hmdb, csp)
```

All the extracted spectra match to the reference spectra from HMDB.

```{r mix04-water-neg-gmp-mirror-hmdb, echo = TRUE, fig.cap = "Mirror plots"}
std_ms2_sel <- plot_select_ms2(std_ms2, std_hmdb, 0.7, ppm = csp@ppm,
                               tolerance = csp@tolerance)
```

We repeat the same with MassBank and we obtain similar results as shown below.

```{r mix04-water-neg-gmp-ms2-mbank, echo = FALSE, fig.width = 5, fig.height = 5}
std_mbank <- get_mbank(mbank, inchikey = unique(std_hmdb$inchikey))
sim <- plot_ms2_similarity_heatmap(std_ms2, std_mbank, csp)
```

```{r mix04-water-neg-gmp-mirror-mbank, echo = TRUE, fig.cap = "Mirror plots"}
tmp_sel <- plot_select_ms2(std_ms2, std_mbank, 0.7, ppm = csp@ppm,
                               tolerance = csp@tolerance)
```

In addition we match (**all**) the MS2 spectra for the matched features against
all spectra from HMDB or MassBank identifying reference spectra with a
similarity larger than 0.7. The results (if any spectra matched) are shown in
the two following tables.

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, hmdb, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, mbank, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

#### Summary

- FT3107 (RT = 225.7, `[M-H]-` ion) with confidence level **A**.
- Reference MS2: `[M-H]-`: FT3107_F15.S0745, FT3107_F16.S0743, FT3107_F16.S0845;
  high confidence.

```{r, echo = FALSE}
## Define the feature (ion) and MS2 spectra to insert
fts <- data.frame(feature_id = c("FT3107"),
                  confidence_level = c("A"))

## Get MS2 spectra:
ms2 <- std_ms2[c("FT3107_F15.S0745", "FT3107_F16.S0743", "FT3107_F16.S0845")]
ms2$confidence <- c(rep("low"), rep())
```

```{r, add-ions, echo = FALSE}
```

```{r, add-ms2-spectra, echo = FALSE}
```


### Leucine

```{r, echo = FALSE}
std <- "Leucine"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

We next plot the EIC for the assigned feature and visually inspect these.

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WN", std_ms2)
```

The cleaned MS2 spectra for the features matched to `r std` are shown below
(peaks with an intensity below 5% of the maximum peak and spectra with less
than 2 peaks were removed).

```{r mix04-water-neg-leucine-ms2, echo = FALSE}
plot_spectra(std_ms2)
```

We next match the extracted MS2 spectra against the reference spectra for 
`r std` from HMDB and MassBank.

```{r mix04-water-neg-leucine-ms2-hmdb, echo = FALSE, fig.width = 5, fig.height = 5}
hmdb_id <- std_dilution$HMDB[std_dilution$name == std]
std_hmdb <- Spectra(cdb, filter = ~ compound_id == hmdb_id)
sim <- plot_ms2_similarity_heatmap(std_ms2, std_hmdb, csp)
```
 
```{r mix04-water-neg-leucine-mirror-hmdb, echo = TRUE, fig.cap = "Mirror plots"}
std_ms2_sel <- plot_select_ms2(std_ms2, std_hmdb, 0.7,
                               ppm = csp@ppm, tolerance = csp@tolerance)
```

We repeat the same with MassBank and we obtain similar results as shown below.

```{r mix04-water-neg-leucine-acid-ms2-mbank, echo = FALSE, fig.width = 5, fig.height = 5}
std_mbank <- get_mbank(mbank, inchikey = unique(std_hmdb$inchikey))
sim <- plot_ms2_similarity_heatmap(std_ms2, std_mbank, csp)
```

```{r mix04-water-neg-leucine-mirror-mbank, echo = TRUE, fig.cap = "Mirror plots"}
tmp_sel <- plot_select_ms2(std_ms2, std_mbank, 0.7,
                           ppm = csp@ppm, tolerance = csp@tolerance)
```

Unfortunately only the precursor m/z is matching.

In addition we match (**all**) the MS2 spectra for the matched features against
all spectra from HMDB or MassBank identifying reference spectra with a
similarity larger than 0.7. The results (if any spectra matched) are shown in
the two following tables.

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, hmdb, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, mbank,
              sv = c("rtime", "target_name", "target_inchikey", "score"),
              name = "target_name", param = csp)
```


#### Summary

- FT0182 (RT = 148.9, `[M-H]-` ion) with confidence level **B**.
- Reference MS2: `[M-H]-`: FT0182_F15.S0455, FT0182_F16.S0478; high confidence.

```{r, echo = FALSE}
## Get MS2 spectra:
ms2 <- std_ms2[c("FT0182_F15.S0455", "FT0182_F16.S0478")]
ms2$confidence <- c("high")
```

```{r, add-ms2-spectra, echo = FALSE}
```


### Pyridoxine


```{r, echo = FALSE}
std <- "Pyridoxine"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

We next plot the EIC for the assigned features and visually inspect these.

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WN", std_ms2)
```

The cleaned MS2 spectra for the features matched to `r std` are shown below
(peaks with an intensity below 5% of the maximum peak and spectra with less
than 2 peaks were removed).

```{r mix04-water-neg-pyridoxine-ms2, echo = FALSE}
plot_spectra(std_ms2)
```

We next match the extracted MS2 spectra against the reference spectra for 
`r std` from HMDB and MassBank.

```{r mix04-water-neg-pyridoxine-ms2-hmdb, echo = FALSE, fig.width = 5, fig.height = 5}
hmdb_id <- std_dilution$HMDB[std_dilution$name == std]
std_hmdb <- Spectra(cdb, filter = ~ compound_id == hmdb_id)
sim <- plot_ms2_similarity_heatmap(std_ms2, std_hmdb, csp)
```

The MS2 spectra from FT0513 match reference spectra from `r std` with high
similarity scores, those from FT0990 with low scores.

```{r mix04-water-neg-pyridoxine-hmdb, echo = TRUE, fig.cap = "Mirror plots"}
std_ms2_sel <- plot_select_ms2(std_ms2, std_hmdb, 0.7, ppm = csp@ppm,
                               tolerance = csp@tolerance)
```

```{r mix04-water-neg-pyridoxine-ms2-mbank, echo = FALSE, fig.width = 5, fig.height = 5}
std_mbank <- get_mbank(mbank, inchikey = unique(std_hmdb$inchikey))
sim <- plot_ms2_similarity_heatmap(std_ms2, std_mbank, csp)
```

In addition we match (**all**) the MS2 spectra for the matched features against
all spectra from HMDB or MassBank identifying reference spectra with a
similarity larger than 0.7. The results (if any spectra matched) are shown in
the two following tables.

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, hmdb, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, mbank, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

#### Summary

- FT0513 (RT = 94.3, `[M-H]-` ion) with confidence level **A**.
- FT0990 (RT = 94.2, `[M+HCOO]-` ion) with confidence level **A-**.
- Reference MS2: `[M-H]-`: FT0513_F15.S0303, FT0513_F16.S0293, FT0513_F16.S0317;
  high confidence. `[M+HCOO]-`: FT0990_F15.S0306, FT0990_F16.S0280,
  FT0990_F16.S0328; low confidence.

```{r, echo = FALSE}
## Define the feature (ion) and MS2 spectra to insert
fts <- data.frame(feature_id = c("FT0513", "FT0990"),
                  confidence_level = c("A", "A-"))

## Get MS2 spectra:
ms2 <- std_ms2[c("FT0513_F15.S0303", "FT0513_F16.S0293", "FT0513_F16.S0317",
                 "FT0990_F15.S0306", "FT0990_F16.S0280", "FT0990_F16.S0328")]
ms2$confidence <- c(rep("high", 3), rep("low", 3))
```

```{r, add-ions, echo = FALSE}
```

```{r, add-ms2-spectra, echo = FALSE}
```


### Riboflavin

```{r, echo = FALSE}
std <- "Riboflavin"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

Many features have been assigned to this standard based on m/z matching. Based
on their retention times, these seem however to represent ions from different
compounds.

We next plot the EIC for the assigned feature and visually inspect these.

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WN", std_ms2)
```

The cleaned MS2 spectra for the features matched to `r std` are shown below
(peaks with an intensity below 5% of the maximum peak and spectra with less
than 2 peaks were removed).

```{r mix04-water-neg-riboflavin-ms2, echo = FALSE}
plot_spectra(std_ms2)
```

We next match the extracted MS2 spectra against the reference spectra for 
`r std` from HMDB and MassBank.

```{r mix04-water-neg-riboflavin-ms2-hmdb, echo = FALSE, fig.width = 5, fig.height = 5}
hmdb_id <- std_dilution$HMDB[std_dilution$name == std]
std_hmdb <- Spectra(cdb, filter = ~ compound_id == hmdb_id)
sim <- plot_ms2_similarity_heatmap(std_ms2, std_hmdb, csp)
```

FT3977, FT3306, FT3847
We find good matches for spectra from FTO950 (especially F15.S0493 but also
F16.S0501).

```{r mix04-water-neg-riboflavin-hmdb, echo = TRUE, fig.cap = "Mirror plots"}
std_ms2_sel <- plot_select_ms2(std_ms2, std_hmdb, 0.7, ppm = csp@ppm,
                               tolerance = csp@tolerance)
```

```{r mix04-water-neg-riboflavin-ms2-mbank, echo = FALSE, fig.width = 5, fig.height = 5}
std_mbank <- get_mbank(mbank, inchikey = unique(std_hmdb$inchikey))
sim <- plot_ms2_similarity_heatmap(std_ms2, std_mbank, csp)
```

In addition we match (**all**) the MS2 spectra for the matched features against
all spectra from HMDB or MassBank identifying reference spectra with a
similarity larger than 0.7. The results (if any spectra matched) are shown in
the two following tables.

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, hmdb, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```

```{r, echo = FALSE, message = FALSE, results = "asis"}
perform_match(std_ms2, mbank, sv = c("rtime", "target_name", "score"),
              name = "target_name", param = csp)
```


#### Summary

- FT3306 (RT = 145.6, `[M-H]-` ion) with confidence **B**.
- FT3977 (RT = 145.8, `[M+HCOO]-` ion) with confidence **B**.
- FT3847 (RT = 145.1, `[M+Cl]-` ion) with confidence **B-**.
- Reference MS2: `[M-H-]`: FT3306_F15.S0457, FT3306_F16.S0467; high confidence.
  `[M+HCOO]-`: FT3977_F15.S0464 , FT3977_F16.S0471; high confidence.
  `[M+Cl]-`: FT3847_F15.S0448, FT3847_F16.S0431; low confidence.

```{r, echo = FALSE}
## Define the feature (ion) and MS2 spectra to insert
fts <- data.frame(feature_id = c("FT3306", "FT3977", "FT3847"),
                  confidence_level = c("B", "B", "B-"))

## Get MS2 spectra:
ms2 <- std_ms2[c("FT3306_F15.S0457", "FT3306_F16.S0467", "FT3977_F15.S0464",
                 "FT3977_F16.S0471", "FT3847_F15.S0448", "FT3847_F16.S0431")]
ms2$confidence <- c(rep("high", 4), rep("low", 2))
```

```{r, add-ions, echo = FALSE}
```

```{r, add-ms2-spectra, echo = FALSE}
```


### SAMe

```{r, echo = FALSE}
std <- "SAMe"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

We next plot the EIC for the assigned feature and visually inspect these.

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WN", std_ms2)
```

The only available MS2 spectrum (cleaned) for the features matched to `r std` is
shown below (peaks with an intensity below 5% of the maximum peak and spectra
with less than 2 peaks were removed).

```{r mix04-water-neg-same-ms2, echo = FALSE}
plot_spectra(std_ms2)
```

#### Summary

- Inconclusive: no MS2 spectra available.
- Eventually FT4153 (RT = 145.1, `[M+Cl]-` ion): confidence level **D**.


### Sedoheptulose-7-phosphate

```{r, echo = FALSE}
std <- "Sedoheptulose-7-phosphate"
```

```{r, table-feature-matches, results = "asis", echo = FALSE, message = FALSE}
```

Many features have been assigned to this standard based on
m/z matching. Based on their retention times, these seem however to represent
ions from different compounds.

We next plot the EIC for the assigned feature and visually inspect these.

```{r, echo = FALSE}
plot_eics(data, std, feature_table, "WN", std_ms2)
```

The cleaned MS2 spectra for the features matched to `r std` are shown below
(peaks with an intensity below 5% of the maximum peak and spectra with less
than 2 peaks were removed).

```{r mix04-water-neg-sedoheptulose-7-phosphate-ms2, echo = FALSE}
plot_spectra(std_ms2)
```

We next match the extracted MS2 spectra against the reference spectra for 
`r std` from HMDB and MassBank.

```{r mix04-water-neg-sedoheptulose-7-phosphate-ms2-hmdb, echo = FALSE, fig.width = 5, fig.height = 5}
hmdb_id <- std_dilution$HMDB[std_dilution$name == std]
std_hmdb <- Spectra(cdb, filter = ~ compound_id == hmdb_id)
sim <- plot_ms2_similarity_heatmap(std_ms2, std_hmdb, csp)
```


```{r mix04-water-neg-sedoheptulose-7-phosphate-ms2-mbank, echo = FALSE, fig.width = 5, fig.height = 5}
std_mbank <- get_mbank(mbank, inchikey = unique(std_hmdb$inchikey))
sim <- plot_ms2_similarity_heatmap(std_ms2, std_mbank, csp)
```

#### Summary

- Inconclusive: no reference spectra available but EIC data looks clear. Need to
  cross-check with positive polarity and serum data.
- FT1816 (RT = 231.3, `[M-H]-` ion): confidence level **D**.
- Reference MS2: `[M-H]-`: FT1816_F15.S0765, FT1816_F15.S0824, FT1816_F16.S0758;
  low confidence.

```{r, echo = FALSE}
## Define the feature (ion) and MS2 spectra to insert
fts <- data.frame(feature_id = c("FT1816"),
                  confidence_level = c("D"))

## Get MS2 spectra:
ms2 <- std_ms2[c("FT1816_F15.S0765", "FT1816_F15.S0824", "FT1816_F16.S0758")]
ms2$confidence <- c("low")
```

```{r, add-ions, echo = FALSE}
```

```{r, add-ms2-spectra, echo = FALSE}
```


## Standards without any matching feature

### Phenylethylamine

### Propylene glycol

### Putrescine

### Ribose

# Summary on the ion database

Summarizing the content that was added to the `IonDb`.

```{r, iondb-summary, echo = FALSE, results = "asis"}
```

# Session information

The R version and packages used in this analysis are listed below.

```{r sessioninfo}
sessionInfo()
```